Tiny block-size coding for energy-efficient image compression and communication in wireless camera sensor networks

This article presents a lightweight image compression algorithm explicitly designed for resource-constrained wireless camera sensors, called TiBS (tiny block-size image coding). TiBS operates on blocks of 2×22×2 pixels (this makes it easy for the end-user to conceal missing blocks due to packet losses) and is based on pixel removal. Furthermore, TiBS is combined with a chaotic pixel mixing scheme to reinforce the robustness of image communication against packet losses. For validation purposes, TiBS as well as a JPEG-like algorithm have been implemented on a real wireless camera sensor composed of a Mica2 mote and a Cyclops imager. The experimental results show that TiBS does not provide high compression ratios, but it enables energy-efficient image communication, even for the source camera node, and even for high packet loss rates. Considering an original 8-bpp grayscale image for instance, the amount of energy consumed by the Cyclops/Mica2 can be reduced by around 60% when the image is compressed using TiBS, compared to the scenario without compression. Moreover, the visual quality of reconstructed images is usually acceptable under packet losses conditions up to 40–50%. In comparison, the JPEG-like algorithm results in clearly more energy consumption than TiBS at similar image quality and, of course, its resilience to packet losses is lower because of the larger size of encoded blocks. Adding redundant packets to the JPEG-encoded data packets may be considered to deal with packet losses, but the energy problem remains.

► A very low complexity algorithm for lossy image compression is proposed.
► It addresses both the problems of energy efficiency and packet loss in wireless camera sensor networks.
► It operates on blocks of 2×2 pixels to make easier the concealment of lost blocks at the decoder.
► The robustness against packet losses is reinforced by a chaotic mixing of the blocks of pixels.
► The results of energy consumption come from measurements on a real wireless camera sensor.